Management of knowledge-based system projects is the focus of this practical book about the development of knowledge-based systems. Part 1 starts with some technical aspects. After a short examination of the basic features of a knowledge-based system, chapters 2 through 5 present the development of four prototype systems that were implemented using existing tools that are available on a personal computer:
a mortgage loan analyzer using 1st-CLASS Fusion,
a financial adviser that recommends an investment strategy using Personal Consultant Easy,
a personnel recruiting adviser using VP-Expert, and
an adviser for knowledge-based system developers using KnowledgePro.
Part 2 presents a six-step approach to knowledge-based system development in six chapters:
selecting a problem,
selling the project,
designing the process,
selecting tools,
implementation, and
fielding the system.
Throughout these steps, the authors stress the need for approval by management, end users, and domain experts. Chapter 12 contains two interviews with experienced people in the field. Three appendices give the reader an overview of periodicals in the field, a number of project ideas, and a partial list of PC tools.
This book does not contain much theory, so it is not useful as a textbook. References to relevant material and a bibliography are also missing. The emphasis is on practical, managerial issues, so people who want to build a system will find a lot of useful guidelines and examples. More than 120 screens illustrate the development and consultation of the prototypes in Part 1, while practical guidelines are presented in numerous steps and bulleted lists in Part 2.
Because of the redundancy of screens and the duplication of lists (which make reading Part 2 very annoying), this book is too long. Chapter 12 could be dropped entirely because it does not add any new information. Also, it was not necessary to spend a whole chapter on each development step. Reading the whole procedure will take days, while the authors are not sure whether a prototype system should be finished in one hour (p. 30) or in two (p. 218).
Typographically, the book has some severe shortcomings. Because of the absence of heading numbers and because an inadequate use of fonts does not allow the reader to make a clear distinction among levels, it takes some time to find the structure of a chapter. Text is sometimes difficult to distinguish from figures, and some examples will be difficult for non–English speakers to understand without additional explanation.
The authors succeed in their basic purpose of teaching the reader to build his or her own system. For this purpose, they provide a step-by-step procedure and a number of examples. The reader will not gain theoretical insight from this book, however.